vellum-ai 1.7.3__tar.gz → 1.7.4__tar.gz

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vellum-ai
-Version: 1.7.3
+Version: 1.7.4
 Summary: 
 License: MIT
 Requires-Python: >=3.9,<4.0

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/pyproject.toml

@@ -3,7 +3,7 @@ name = "vellum-ai"
 
 [tool.poetry]
 name = "vellum-ai"
-version = "1.7.3"
+version = "1.7.4"
 description = ""
 readme = "README.md"
 authors = []

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/client/core/client_wrapper.py

@@ -27,10 +27,10 @@ class BaseClientWrapper:
 
     def get_headers(self) -> typing.Dict[str, str]:
         headers: typing.Dict[str, str] = {
-            "User-Agent": "vellum-ai/1.7.3",
+            "User-Agent": "vellum-ai/1.7.4",
             "X-Fern-Language": "Python",
             "X-Fern-SDK-Name": "vellum-ai",
-            "X-Fern-SDK-Version": "1.7.3",
+            "X-Fern-SDK-Version": "1.7.4",
             **(self.get_custom_headers() or {}),
         }
         if self._api_version is not None:

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/client/core/pydantic_utilities.py

@@ -158,7 +158,14 @@ class UniversalBaseModel(pydantic.BaseModel):
 
             dict_dump = super().dict(**kwargs_with_defaults_exclude_unset_include_fields)
 
-        return convert_and_respect_annotation_metadata(object_=dict_dump, annotation=self.__class__, direction="write")
+        # Skip expensive annotation metadata conversion for types that don't require it.
+        # This optimization dramatically improves serialization performance for large nested
+        # structures by avoiding repeated typing.get_origin checks on every element.
+        if self.__class__.__name__ in annotated_types:
+            return convert_and_respect_annotation_metadata(
+                object_=dict_dump, annotation=self.__class__, direction="write"
+            )
+        return dict_dump
 
 
 def _union_list_of_pydantic_dicts(source: List[Any], destination: List[Any]) -> List[Any]:

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/plugins/pydantic.py

@@ -48,6 +48,7 @@ class OnValidatePython(ValidatePythonHandlerProtocol):
         from_attributes: Optional[bool] = None,
         context: Optional[Dict[str, Any]] = None,
         self_instance: Optional[Any] = None,
+        extra: Optional[Literal["allow", "ignore", "forbid"]] = None,
         allow_partial: Union[bool, Literal["off", "on", "trailing-strings"]] = False,
         by_alias: Optional[bool] = None,
         by_name: Optional[bool] = None,

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/workflows/errors/types.py

@@ -33,6 +33,7 @@ class WorkflowError:
     message: str
     code: WorkflowErrorCode
     raw_data: Optional[Dict[str, Any]] = None
+    stacktrace: Optional[str] = None
 
     def __contains__(self, item: Any) -> bool:
         return item in self.message
@@ -82,6 +83,8 @@ def workflow_event_error_to_workflow_error(error: WorkflowEventError) -> Workflo
     return WorkflowError(
         message=error.message,
         code=_WORKFLOW_EVENT_ERROR_CODE_TO_WORKFLOW_ERROR_CODE.get(error.code, WorkflowErrorCode.INTERNAL_ERROR),
+        raw_data=error.raw_data,
+        stacktrace=error.stacktrace,
     )
 
 

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/workflows/events/tests/test_event.py

@@ -263,6 +263,7 @@ mock_node_uuid = str(uuid4_from_hash(MockNode.__qualname__))
                         "message": "Workflow failed",
                         "code": "USER_DEFINED_ERROR",
                         "raw_data": None,
+                        "stacktrace": None,
                     },
                     "stacktrace": None,
                 },

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/workflows/exceptions.py

@@ -25,10 +25,12 @@ class NodeException(Exception):
         message: str,
         code: WorkflowErrorCode = WorkflowErrorCode.INTERNAL_ERROR,
         raw_data: Optional[Dict[str, Any]] = None,
+        stacktrace: Optional[str] = None,
     ):
         self.message = message
         self.code = code
         self.raw_data = raw_data
+        self.stacktrace = stacktrace
         super().__init__(message)
 
     @property
@@ -37,11 +39,17 @@ class NodeException(Exception):
             message=self.message,
             code=self.code,
             raw_data=self.raw_data,
+            stacktrace=self.stacktrace,
         )
 
     @staticmethod
     def of(workflow_error: WorkflowError) -> "NodeException":
-        return NodeException(message=workflow_error.message, code=workflow_error.code, raw_data=workflow_error.raw_data)
+        return NodeException(
+            message=workflow_error.message,
+            code=workflow_error.code,
+            raw_data=workflow_error.raw_data,
+            stacktrace=workflow_error.stacktrace,
+        )
 
 
 class WorkflowInitializationException(Exception):

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/workflows/nodes/displayable/subworkflow_deployment_node/node.py

@@ -21,7 +21,7 @@ from vellum.workflows.errors import WorkflowErrorCode
 from vellum.workflows.errors.types import workflow_event_error_to_workflow_error
 from vellum.workflows.events.types import default_serializer
 from vellum.workflows.events.workflow import is_workflow_event
-from vellum.workflows.exceptions import NodeException
+from vellum.workflows.exceptions import NodeException, WorkflowInitializationException
 from vellum.workflows.inputs.base import BaseInputs
 from vellum.workflows.nodes.bases.base import BaseNode
 from vellum.workflows.outputs.base import BaseOutput
@@ -164,6 +164,16 @@ class SubworkflowDeploymentNode(BaseNode[StateType], Generic[StateType]):
                 node_output_mocks=self._context._get_all_node_output_mocks(),
             )
 
+        try:
+            next(subworkflow_stream)
+        except WorkflowInitializationException as e:
+            hashed_module = e.definition.__module__
+            raise NodeException(
+                message=e.message,
+                code=e.code,
+                raw_data={"hashed_module": hashed_module},
+            ) from e
+
         outputs = None
         exception = None
         fulfilled_output_names: Set[str] = set()

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/workflows/runner/runner.py

@@ -1,4 +1,5 @@
 from collections import defaultdict
+from contextlib import contextmanager
 from copy import deepcopy
 from dataclasses import dataclass
 import logging
@@ -232,7 +233,33 @@ class WorkflowRunner(Generic[StateType]):
 
         self._background_thread: Optional[Thread] = None
         self._cancel_thread: Optional[Thread] = None
-        self._stream_thread: Optional[Thread] = None
+
+    @contextmanager
+    def _httpx_logger_with_span_id(self) -> Iterator[None]:
+        """
+        Context manager to append the current execution's span ID to httpx logger messages.
+
+        This is used when making API requests via the Vellum client to include
+        the execution's span ID in the httpx request logs for better traceability.
+        """
+        httpx_logger = logging.getLogger("httpx")
+
+        class SpanIdFilter(logging.Filter):
+            def filter(self, record: logging.LogRecord) -> bool:
+                if record.name == "httpx" and "[span_id=" not in record.msg:
+                    context = get_execution_context()
+                    if context.parent_context:
+                        span_id = str(context.parent_context.span_id)
+                        record.msg = f"{record.msg} [span_id={span_id}]"
+                return True
+
+        span_filter = SpanIdFilter()
+        httpx_logger.addFilter(span_filter)
+
+        try:
+            yield
+        finally:
+            httpx_logger.removeFilter(span_filter)
 
     def _snapshot_state(self, state: StateType, deltas: List[StateDelta]) -> StateType:
         self._workflow_event_inner_queue.put(
@@ -707,6 +734,7 @@ class WorkflowRunner(Generic[StateType]):
             error_message: The error message to include in the cancellation events
             parent_context: The parent context for the cancellation events
         """
+        captured_stacktrace = "".join(traceback.format_stack())
         for span_id, active_node in list(self._active_nodes_by_execution_id.items()):
             rejection_event = NodeExecutionRejectedEvent(
                 trace_id=self._execution_context.trace_id,
@@ -717,6 +745,7 @@ class WorkflowRunner(Generic[StateType]):
                         code=WorkflowErrorCode.NODE_CANCELLED,
                         message=error_message,
                     ),
+                    stacktrace=captured_stacktrace,
                 ),
                 parent=parent_context,
             )
@@ -824,6 +853,11 @@ class WorkflowRunner(Generic[StateType]):
             parent=self._execution_context.parent_context,
         )
 
+    def _wrapped_stream(self) -> None:
+        """Wrapper for _stream that adds httpx logger span ID context."""
+        with self._httpx_logger_with_span_id():
+            self._stream()
+
     def _stream(self) -> None:
         for edge in self.workflow.get_edges():
             self._dependencies[edge.to_node].add(edge.from_port.node_class)
@@ -960,12 +994,14 @@ class WorkflowRunner(Generic[StateType]):
                     parent_context=parent_context,
                 )
 
+                captured_stacktrace = "".join(traceback.format_stack())
                 self._workflow_event_outer_queue.put(
                     self._reject_workflow_event(
                         WorkflowError(
                             code=WorkflowErrorCode.WORKFLOW_CANCELLED,
                             message="Workflow run cancelled",
-                        )
+                        ),
+                        captured_stacktrace,
                     )
                 )
                 return
@@ -1005,7 +1041,7 @@ class WorkflowRunner(Generic[StateType]):
 
         # The extra level of indirection prevents the runner from waiting on the caller to consume the event stream
         self._stream_thread = Thread(
-            target=self._stream,
+            target=self._wrapped_stream,
             name=f"{self.workflow.__class__.__name__}.stream_thread",
         )
         self._stream_thread.start()

{vellum_ai-1.7.3 → vellum_ai-1.7.4}/src/vellum/workflows/types/core.py

@@ -1,4 +1,5 @@
 from enum import Enum
+from multiprocessing.synchronize import Event as MultiprocessingEvent
 from threading import Event as ThreadingEvent
 from typing import (  # type: ignore[attr-defined]
     Any,
@@ -16,7 +17,7 @@ JsonArray = List["Json"]
 JsonObject = Dict[str, "Json"]
 Json = Union[None, bool, int, float, str, JsonArray, JsonObject]
 
-CancelSignal = ThreadingEvent
+CancelSignal = Union[ThreadingEvent, MultiprocessingEvent]
 
 # Unions and Generics inherit from `_GenericAlias` instead of `type`
 # In future versions of python, we'll see `_UnionGenericAlias`